Synthetic ammonia production control method, system, equipment, medium and product

Abstract

The invention relates to the technical field of industrial control, in particular to a synthetic ammonia production control method, system and equipment, a medium and a product. The method comprises the following steps: acquiring production modeling data of a synthetic ammonia field section, performing unsupervised analysis on the production modeling data through a preset big data model, determining key features and an analysis result of the key features in a synthetic ammonia production process, performing visualization processing on the key features, generating a visual chart, and displaying the visual chart on a database. Therefore, semantic analysis and causal reasoning are carried out on the visual chart and the analysis result through a preset large language model, a target optimization strategy is generated, optimization control is carried out on a synthetic ammonia production site, and the production efficiency of synthetic ammonia is improved.

Description

Technical Field

[0001] This application relates to the field of industrial control technology, and in particular to a method, system, equipment, medium and product for controlling the production of synthetic ammonia. Background Technology

[0002] With the continuous advancement of industrial intelligence, the stability, economy, and safety of the production process of synthetic ammonia, as an important chemical raw material, have a significant impact on the national economy. The synthetic ammonia production process is complex, involving key steps such as high-temperature and high-pressure reactions and the conversion of multiple components, making real-time monitoring and intelligent control increasingly urgent.

[0003] Currently, traditional methods for monitoring synthetic ammonia production rely heavily on decentralized instrument monitoring and manual experience analysis. Their early warning mechanisms, based on a single threshold, struggle to identify potential risks associated with multiple parameter anomalies. Furthermore, their optimization relies on historical experience, making it difficult to adapt to dynamic operating conditions such as raw material fluctuations and equipment aging. This results in significant fluctuations in synthetic ammonia production efficiency, high energy consumption, and reduced overall production efficiency.

[0004] Therefore, there is an urgent need for a method to control the production of synthetic ammonia in order to improve the efficiency of synthetic ammonia production. Summary of the Invention

[0005] This invention provides a method, equipment, storage medium, and program product for controlling ammonia synthesis production, thereby improving the accuracy and efficiency of assessing the impact of unplanned shutdowns of ammonia synthesis plant production units.

[0006] In a first aspect, this application provides a method for controlling the production of synthetic ammonia, the method comprising:

[0007] Acquire production modeling data for the ammonia synthesis site, wherein the production modeling data characterizes the structured process parameters and site configuration information during the ammonia synthesis production process;

[0008] Based on a pre-defined big data model, unsupervised analysis is performed on the production modeling data to determine the key features in the ammonia synthesis process and the analysis results of the key features; the key features characterize the parameters affecting ammonia production and energy consumption, and the analysis results characterize the anomaly identification, energy consumption analysis, and production prediction results of each parameter;

[0009] The key features are visualized to generate visualization charts; the visualization charts represent the operating status and working condition boundaries of the ammonia synthesis production process.

[0010] Based on a pre-defined large language model, semantic parsing and causal reasoning are performed on the visualization charts and analysis results to generate target optimization strategies.

[0011] Based on the aforementioned target optimization strategy, the synthetic ammonia production site is optimized and controlled.

[0012] Secondly, this application provides a synthetic ammonia production control system, the system comprising:

[0013] The data preprocessing module is used to acquire raw data from the ammonia synthesis field section and preprocess the raw data to obtain production modeling data; the production modeling data represents the structured process parameters and section configuration information in the ammonia synthesis production process.

[0014] The big data feature analysis module is used to perform unsupervised analysis on the production modeling data based on a preset big data model, to determine the key features in the ammonia synthesis process and the analysis results of the key features; the key features characterize the parameters affecting ammonia production and energy consumption, and the analysis results characterize the anomaly identification, energy consumption analysis and production prediction results of each parameter;

[0015] The visualization module is used to visualize the key features and generate visualization charts; the visualization charts represent the operating status and operating condition boundaries of the ammonia synthesis production process.

[0016] The real-time analysis and strategy generation module is used to perform semantic parsing and causal reasoning on the visualization charts and analysis results based on a preset large language model, and generate target optimization strategies.

[0017] The control strategy output module is used to optimize and control the ammonia synthesis production site based on the target optimization strategy.

[0018] Optionally, the data preprocessing module is specifically used for:

[0019] Obtain the raw data of the ammonia synthesis field section, and identify and correct any abnormal missing values ​​in the raw data;

[0020] The processed raw data is then standardized to obtain raw data with a uniform format.

[0021] Based on a preset data integration strategy, the multi-segment data in the original data are correlated and integrated to obtain structured production modeling data.

[0022] Optionally, the big data feature analysis module is specifically used for:

[0023] Based on a preset unsupervised clustering strategy, pattern recognition is performed on the production modeling data to obtain production datasets corresponding to multiple production condition categories.

[0024] Based on a preset feature selection strategy, relevant key features are selected from each production dataset, and each key feature is transformed based on a preset dimensionality reduction strategy to obtain the dimensionality-reduced key features.

[0025] Based on a preset time series analysis strategy, patterns are mined from each key feature to generate corresponding feature evolution curves; the feature evolution curves characterize the dynamic changes of the corresponding key features with the production process.

[0026] Optionally, the big data model is constructed through the following steps:

[0027] Based on a preset dimensionality reduction strategy, the production modeling data is dimensionality reduced to obtain a training dataset.

[0028] Based on a preset unsupervised learning strategy and time series analysis strategy, the model is trained on the training dataset to obtain the big data model.

[0029] Optionally, the visualization chart includes at least one of a big data multidimensional pattern diagram, a variable trend diagram, and a safety analysis list; the big data multidimensional pattern diagram represents the collaborative relationship and safety boundary of the operating status of each section in the ammonia synthesis process; the variable trend diagram represents the fluctuation pattern of each production parameter over time in the ammonia synthesis process; and the safety analysis list includes quantitative safety information and abnormal early warning information of the ammonia synthesis production status.

[0030] Optionally, the control strategy output module is specifically used for:

[0031] Based on the aforementioned target optimization strategy, the production equipment at the ammonia synthesis production site is controlled to perform corresponding operations.

[0032] The production modeling data is updated based on the operation results.

[0033] Thirdly, this application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the synthetic ammonia production control methods described in the first aspect above.

[0034] Fourthly, this application provides a computer storage medium storing computer program instructions, which are executed by a processor using any of the synthetic ammonia production control methods described in the first aspect above.

[0035] Fifthly, an embodiment of this application provides a computer program product including computer program instructions, which, when executed by a processor, implement any one of the synthetic ammonia production control methods described in the first aspect above.

[0036] The beneficial effects of this invention are as follows:

[0037] This application provides a method, system, equipment, medium, and product for controlling the production of synthetic ammonia. The method acquires production modeling data of the synthetic ammonia production site and performs unsupervised analysis on the production modeling data through a preset big data model to determine the key features and analysis results of the key features in the synthetic ammonia production process. The key features are then visualized to generate visual charts. Subsequently, a preset big language model is used to perform semantic parsing and causal reasoning on the visual charts and analysis results to generate target optimization strategies for optimizing and controlling the synthetic ammonia production site and improving the efficiency of synthetic ammonia production. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0039] Figure 1 A schematic diagram of a synthetic ammonia production control process provided for an embodiment of this application;

[0040] Figure 2 A schematic diagram of a security analysis list provided in an embodiment of this application;

[0041] Figure 3 A schematic diagram illustrating the generation process of a target optimization strategy provided in an embodiment of this application;

[0042] Figure 4 This is a schematic diagram of the structure of a synthetic ammonia production control system provided in an embodiment of this application;

[0043] Figure 5 This is a schematic diagram of the structure of a data preprocessing module provided in an embodiment of this application;

[0044] Figure 6 A schematic diagram illustrating the workflow of a big data feature analysis module provided in this application embodiment;

[0045] Figure 7 A schematic diagram illustrating the workflow of a visualization module provided in an embodiment of this application;

[0046] Figure 8 A flowchart illustrating a real-time analysis and strategy generation module provided in an embodiment of this application;

[0047] Figure 9 This is a schematic diagram of the control process of a target optimization strategy provided in an embodiment of this application;

[0048] Figure 10This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Unless otherwise specified, the embodiments and features in the embodiments of this application can be arbitrarily combined with each other. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0050] The terms "first" and "second" in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the term "comprising" and any variations thereof are intended to cover non-exclusive protection. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices. The term "multiple" in this application can mean at least two, for example, two, three, or more, and this application does not impose limitations.

[0051] The term "and / or" in the embodiments of this application is merely a description of the association relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0052] It is understood that the following specific embodiments of this application involve data related to synthetic ammonia production. When the various embodiments of this application are applied to specific products or technologies, relevant licenses or consents are required, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, relevant volunteers can be recruited and agreements can be signed to authorize their data, thereby enabling the implementation using the data of these volunteers; alternatively, implementation can be carried out within an authorized organization, using data from members of the organization to implement the following implementation methods for data management; or, in specific implementations, the relevant data used are all simulated data, such as simulated data generated in a virtual scenario.

[0053] The design concept of the embodiments of this application is briefly introduced below:

[0054] With the continuous advancement of industrial intelligence, synthetic ammonia, as an important chemical raw material, involves complex processes such as high-temperature and high-pressure reactions, multi-component material conversion, and multi-stage collaboration in its production process. Production stability, energy consumption control, and safety management directly affect the economic benefits and industry competitiveness of enterprises.

[0055] Currently, traditional ammonia synthesis production control typically relies on decentralized instrument monitoring and manual experience-based decision-making. Monitoring of key process parameters is primarily achieved through single-point instruments, and production data analysis depends on offline processing and expert judgment by engineers. However, this traditional approach has significant limitations: Firstly, data from each stage is stored in independent systems, lacking unified integration and standardized processing, making it difficult to achieve end-to-end correlation analysis, especially when facing complex operating conditions with multiple coupled variables, resulting in sluggish responses. Secondly, the identification of ammonia synthesis production anomalies largely depends on operator inspections and historical experience. When facing nonlinear operating conditions with multiple coupled variables (e.g., efficiency decline due to fluctuations in raw coal quality, and a decrease in ammonia net value due to catalyst activity decay), not only is the response sluggish, but potential risks are easily missed due to human judgment bias. Furthermore, energy optimization strategies are mostly based on fixed process parameters, making it difficult to adapt to dynamic production conditions, leading to energy waste and capacity fluctuations. For example, traditional ammonia synthesis production monitoring interfaces are mainly static reports, failing to intuitively present the dynamic correlations between multiple stages (e.g., the coordinated changes in energy consumption and output), resulting in a heavy cognitive burden and sluggish responses for operators. Early warning mechanisms in ammonia synthesis production typically rely on single threshold judgments, failing to identify multi-parameter synergistic anomalies (e.g., the coupling risk of raw material fluctuations and equipment aging), and easily missing potential safety incidents. Furthermore, the control and optimization of ammonia synthesis production processes depend on engineers' historical experience, making it difficult to adapt to dynamic operating conditions (e.g., fluctuations in raw coal quality and catalyst activity decay), leading to large fluctuations in production efficiency and high energy consumption.

[0056] In view of the above problems, embodiments of this application provide a method, system, equipment, medium, and product for controlling synthetic ammonia production. The method acquires production modeling data of the synthetic ammonia production site, performs unsupervised analysis on the production modeling data through a preset big data model, determines the key features and analysis results of the key features in the synthetic ammonia production process, and visualizes the key features to generate visual charts. Then, through a preset big language model, semantic parsing and causal reasoning are performed on the visual charts and analysis results to generate target optimization strategies for optimizing and controlling the synthetic ammonia production site and improving synthetic ammonia production efficiency.

[0057] Furthermore, this application also provides a synthetic ammonia production control system. This system standardizes various raw data from the synthetic ammonia production process to obtain structured production modeling data. Based on a big data model, it performs unsupervised analysis on the production modeling data to extract key features. The system then visualizes the production status and sets target variables. Next, it uses the big data model to analyze the on-site operating conditions in real time, generates target optimization strategies based on a large language model, and controls the production site according to these strategies. This approach integrates and organizes multi-source data, effectively improving the quality of modeling data. Through big data analysis and visualization technology, it achieves comprehensive perception of the production status. The large language model enhances the intelligence and interpretability of the optimization strategies, significantly improving the monitoring accuracy, early warning timeliness, and production efficiency of the synthetic ammonia production process, reducing energy consumption fluctuations, and achieving intelligent management and precise optimization throughout the entire process.

[0058] Furthermore, the synthetic ammonia production control system provided in this application significantly addresses the pain points of traditional synthetic ammonia production management through a modular, end-to-end architecture. On one hand, the data preprocessing module achieves standardized integration of multi-source data, eliminating data silos and filtering noisy data to improve the quality of the analysis. On the other hand, the visualization module reduces the cognitive burden on operators and engineers through multi-dimensional correlation, enabling a comprehensive view on a single screen. Moreover, the real-time analysis and strategy generation module, by integrating the advantages of big data models and large language models, ensures both the real-time nature and accuracy of the analysis while improving interpretability through semantic strategies. The control strategy output module achieves a closed loop of "analysis-decision-execution," ensuring the rapid implementation of optimization strategies. Simultaneously, in the event of system failure or network disconnection, each module can retain historical data and the current operating context, eliminating the need for re-initialization upon restart, thus enhancing system fault tolerance and continuity. This application's embodiments, through a modular collaborative architecture and a full-process data management mechanism, achieve multi-dimensional key technical effects in the synthetic ammonia production field, including: an intuitive and efficient production management visualization interface; the visualization module, through multi-format presentation technology, transforms production data into interpretable information, significantly improving operators' overall understanding of production conditions and drastically shortening problem localization time; improved production anomaly identification efficiency; the real-time analysis and strategy generation module integrates process mechanism models and machine learning algorithms to construct a multi-dimensional anomaly identification system, proactively preventing potential safety situations, significantly reducing production safety accident rates and false alarm rates, while effectively reducing production shutdowns; and the generation of safety... With an optimization strategy that balances efficiency and control, the control strategy output module, through the collaboration of a large language model and a process mechanism model, optimizes key variables while ensuring all process parameters meet safety thresholds, effectively reducing energy consumption and increasing synthetic ammonia production. The closed-loop execution of the control strategy output module enables precise control under dynamic operating conditions, reducing the number of manual interventions and minimizing capacity fluctuations. Furthermore, this system not only enhances the intelligence level of synthetic ammonia production but also provides a standardized architecture for industrial continuous production process control. It can be further expanded into multi-stage collaborative optimization and multimodal industrial intelligent systems in the future, becoming an important technological foundation for the intelligent transformation of the synthetic ammonia industry and even the chemical industry.

[0059] The method provided by exemplary embodiments of this application will now be described with reference to the accompanying drawings. It should be noted that the above application scenarios are shown only to facilitate understanding of the spirit and principles of this application, and the embodiments of this application are not limited in any way.

[0060] See Figure 1 The diagram shown is a flowchart of a synthetic ammonia production control method provided in an embodiment of this application. The specific implementation process of this method is as follows:

[0061] Step 101: Obtain production modeling data for the ammonia synthesis field section.

[0062] In this embodiment, the production modeling data of the ammonia synthesis field section represents the structured process parameters and section configuration information in the ammonia synthesis production process, such as variable identifiers, data types, acquisition frequency, measurement range, section affiliation, and data units.

[0063] In one possible implementation, the present application embodiment will acquire the raw data of the synthetic ammonia field section, identify and correct abnormal missing values ​​in the raw data, and then standardize the processed raw data to obtain raw data with a uniform format. Thus, through a preset data integration strategy, the multi-section data in the raw data will be associated and integrated to obtain structured production modeling data.

[0064] Specifically, this application embodiment will collect raw data from the ammonia synthesis field, and then process, clean, and standardize it. The raw data may include unstructured or semi-structured data such as real-time flow, temperature, and pressure collected by sensors during the ammonia synthesis production process, as well as information from various points, such as the corresponding work section, upper and lower thresholds, and set values. In the processing stage, data from different sources and formats will be categorized and integrated; the cleaning stage will remove outliers caused by sensor drift and noise data such as instantaneous fluctuations during equipment start-up and shutdown; the standardization process will unify the data format, units, and acquisition frequency, ultimately obtaining structured production modeling data, laying the foundation for subsequent analysis.

[0065] Step 102: Perform unsupervised analysis on the production modeling data based on the preset big data model to determine the key features in the ammonia synthesis production process and the analysis results of each key feature.

[0066] In this embodiment, unsupervised analysis of production modeling data is performed using big data models such as clustering algorithms and time series analysis to extract key features. At the same time, features are constructed in conjunction with the process mechanism to determine target optimization variables (e.g., core parameters affecting ammonia production and key indicators affecting energy consumption). In addition, a correlation model between key features is constructed in conjunction with the ammonia synthesis process mechanism to more accurately reflect the variable relationships in the production process.

[0067] Specifically, in this application embodiment, unsupervised analysis of production modeling data is performed using a big data model, and key features are extracted, such as the synthesis tower temperature, pressure, and hydrogen-nitrogen ratio that affect ammonia production, and key process parameters such as coal consumption in the gasifier section and electricity consumption in the air separation section that affect energy consumption.

[0068] In one possible implementation, this embodiment of the application will perform pattern recognition on production modeling data according to a preset unsupervised clustering strategy to obtain production datasets corresponding to multiple production condition categories. Then, according to a preset feature selection strategy, corresponding key features will be selected from each production dataset. Finally, each key feature will be transformed according to a preset dimensionality reduction strategy to obtain dimensionality-reduced key features. Next, according to a preset time series analysis strategy, the change patterns of each key feature will be analyzed to generate feature evolution curves. These feature evolution curves represent the dynamic changes of key features with the production process.

[0069] Specifically, this application embodiment can employ unsupervised clustering algorithms such as K-means clustering to aggregate data samples with similar characteristics into the same category by calculating similarity measures between data points, such as Euclidean distance or cosine similarity. This identifies different production operating conditions and groups the production modeling data obtained during the ammonia synthesis production process. In this way, without relying on manually labeled data, hidden regularities can be discovered in the data, outputting production datasets corresponding to multiple production operating condition categories. Each dataset represents a typical production state, providing a classification basis for subsequent feature extraction. Next, using a preset feature selection strategy, such as algorithms based on information gain or recursive feature elimination, variables that significantly affect the ammonia synthesis production state are selected from each dataset, including temperature, pressure, or energy consumption indicators affecting ammonia production, to eliminate redundant variables and focus on core factors. Then, using a preset dimensionality reduction strategy, such as principal component analysis or linear discriminant analysis, the selected high-dimensional data is transformed. By retaining the principle of maximum variance or minimum reconstruction error, the original variables are mapped to a low-dimensional space, extracting principal components or key features that can summarize the essential characteristics of the data. This approach not only reduces data dimensionality and computational complexity but also enhances the robustness and interpretability of features. It identifies a set of key features for each production dataset, reflecting the core attributes of that work condition category and providing purified input for subsequent time series analysis. Using the identified key features, this application employs time series analysis algorithms such as autoregressive integral moving average models or dynamic time warping algorithms to mine patterns in the time-series data of each key feature during the production process. By identifying the trend, periodic, and random components of the data, a dynamic model of feature value changes over time is constructed, generating corresponding feature evolution curves.

[0070] Specifically, this feature evolution curve can be plotted with time on the horizontal axis and feature values ​​on the vertical axis, visually demonstrating the fluctuation patterns of key features as the production process progresses, such as the periodic increase of temperature parameters or the gradual change of pressure indicators. In this way, it not only shows the dynamic evolution trajectory of the features but also reveals the lag correlations and synergistic change patterns between variables, providing a quantitative basis for real-time monitoring and early warning of anomalies in subsequent ammonia synthesis production.

[0071] In one possible implementation, embodiments of this application can use a preset dimensionality reduction strategy to reduce the dimensionality of production modeling data to obtain a training dataset, and then use a preset unsupervised learning strategy and time series analysis strategy to train a model on the training dataset to obtain a big data model.

[0072] Specifically, in this embodiment, the modeling process of the big data model includes two main operations: the application of dimensionality reduction algorithms and model training. Regarding dimensionality reduction algorithms, Fourier transform, wavelet transform, principal component analysis, and K-means clustering can be used. For example, principal component analysis can be used to reduce the dimensionality of numerous related variables in the production modeling data, extracting the principal components that best reflect the data characteristics, reducing data dimensionality while retaining key information. Alternatively, this application can also use K-means clustering to group production data samples with similar characteristics into one class, facilitating subsequent analysis of whether different categories of data exhibit characteristic patterns under different working conditions.

[0073] Furthermore, regarding model training, this application allows for single-model training, fine-tuning, ensemble learning, and iterative optimization. For example, a single machine learning model can be trained first to initially learn the relationship between features in production modeling data and the state of ammonia synthesis production. This model can then be fine-tuned to better reflect actual production conditions. Alternatively, ensemble learning methods can be used to combine multiple different models, improving the accuracy and robustness of the model's analysis of ammonia synthesis production characteristics. In addition, this application can continuously adjust model parameters through iterative optimization, gradually improving model performance. Specifically, considering the wide variety of dimensionality reduction and training algorithms, this application prioritizes unsupervised learning and time-series related technologies, selecting different algorithms and training methods within these categories to achieve an ideal big data model.

[0074] Specifically, taking the production modeling data, which includes 9 datasets in total (8 work sections and 1 overall dataset), as an example, this application can perform feature extraction and model training on the datasets of different work sections separately to obtain big data models for each work section. Simultaneously, a similar operation is performed on the overall factory dataset to obtain a full-process big data model. Thus, the big data model of this embodiment can be adapted to engineers and operators in different work sections, facilitating subsequent visualization modules. Furthermore, the full-process model obtained through the above steps can analyze and monitor the overall operating status of the synthetic ammonia plant. After modeling is completed, real-time field data can be input into the constructed big data model. This real-time field data is data collected in real-time during the synthetic ammonia production process, reflecting the latest production conditions. The number of models can be determined based on the actual situation of the factory. For example, if the number of equipment points in the factory is large, it can be divided into multiple models to avoid an imbalance in variable weights due to an excessive number of variables, which could affect the accuracy of analysis and detection. At the same time, big data model calculations rely on certain hardware computing power requirements. Therefore, the number of models, algorithm and training complexity should be reasonably allocated, and the priority of different models should be appropriately considered based on the actual facilities of the factory. Furthermore, the selection criteria for relevant modeling models should follow statistical indicators, including accuracy, interpretability, and generalization ability, which will not be elaborated upon here.

[0075] Specifically, in this embodiment, big data feature analysis involves training production modeling data using relevant statistical algorithms to obtain key features influencing ammonia synthesis production. A wide variety of big data algorithms are available, and the best one can be selected based on model testing results. However, in general, unsupervised learning algorithms and time series analysis should be prioritized. In this embodiment, unsupervised algorithms do not rely on manually labeled data and can directly discover hidden patterns and correlations from production modeling data, making them suitable for mining complex variable relationships in ammonia synthesis production. Furthermore, ammonia synthesis production data has strong time-series characteristics; point variables change continuously over time and exhibit lagged correlations. Time series analysis techniques can capture the dynamic trends of variables and their dependencies over time. Therefore, this application can select relevant algorithms from unsupervised learning and time series analysis, dividing the production modeling data into training, testing, and validation sets to construct a big data model.

[0076] Step 103: Visualize the key features to generate a visualization chart.

[0077] In this embodiment of the application, by visualizing the key features of the ammonia synthesis process, the production status of ammonia synthesis can be reflected in real time, and real-time visualization of production can be achieved by continuously updating data and charts.

[0078] In one possible implementation, the various forms of visualization charts in this application include, but are not limited to: big data multidimensional pattern diagrams, variable trend diagrams, and safety analysis lists. The big data multidimensional pattern diagram represents the collaborative relationship and safety boundary of the operating status of each section in the ammonia synthesis process, the variable trend diagram represents the fluctuation pattern of each production parameter over time in the ammonia synthesis process, and the safety analysis list represents the safety quantification information and abnormal warning information of the ammonia synthesis production status.

[0079] Specifically, the big data multidimensional pattern diagram in this application refers to reducing all points to a single plane using an unsupervised algorithm model, and constructing boundaries based on relevant statistical intervals, such as confidence intervals, for the points on this plane. Furthermore, a clustering algorithm is used to divide the reduced-dimensional points into different clusters representing different operating conditions. For example, the latest points collected on-site are mapped onto the pattern diagram using the big data model, and their location represents the most probable operating condition. Taking the boiler section pattern diagram as an example, its target variable is thermal efficiency. Historical data visualization shows that clusters with high thermal efficiency are mostly located on the right side of the pattern diagram. If a new mapped point is on the right side of the pattern diagram, it is highly probable that the current operating condition has high thermal efficiency; if a new mapped point is on the left side of the pattern diagram, it represents a high probability that the current operating condition has low thermal efficiency. Simultaneously, if a new mapped point exceeds or approaches the defined boundaries, it most likely indicates a warning or alarm risk. Only in rare cases does exceeding the boundaries represent the emergence of a new production mode.

[0080] Specifically, the variable trend graph in this application embodiment presents the trend of variable change over time, which can more clearly reflect the dynamic characteristics of data in the ammonia synthesis production process and help relevant personnel quickly grasp the patterns and anomalies in the production data.

[0081] Specifically, the security analysis list in this application embodiment displays information such as time, status, and security index, and is updated in real time.

[0082] For details, please refer to Figure 2The diagram shown illustrates a safety analysis list provided in this application embodiment. This list displays the measured values, set values, units, and thresholds for each point (e.g., FI_001, PI_002, TI_003). For example, at 01:00:00 on January 1, 2025, the status is normal, and the safety index is 52.67%. The measured value of PI_002 is 0.85 MPa, the set value is 0.9 MPa, and the thresholds are 0.95 / 0.8 MPa, etc., facilitating a direct view of the safety-related data for each point. It is worth mentioning that this application embodiment demonstrates three types of visualization charts, but this application can also select other various visualization tools based on the actual situation of the plant to intuitively reflect the results of the big data model and the current production status. Through the above-mentioned multi-form visualization, the data in the ammonia synthesis production process becomes more intuitive and easier to understand, providing a clear data presentation for subsequent real-time analysis and strategy generation, and helping to achieve efficient monitoring and analysis of ammonia synthesis production.

[0083] Step 104: Based on the preset large language model, perform semantic parsing and causal reasoning on the visualization charts and analysis results to generate target optimization strategies.

[0084] In this embodiment, the analysis results of the big data model and the visualized charts after visualization processing can be input into the big language model. By combining the big language model with the process knowledge and historical experience of ammonia synthesis production, the input data can be understood and reasoned to generate target optimization strategies to guide the control and operation of the ammonia synthesis production site, so as to achieve optimization, safety and stability control of the production process.

[0085] Specifically, the safety analysis, stability analysis, and optimization analysis in this application can be considered in the order of production safety, production stability, and production optimization, respectively, based on big data models and visualization results, progressing step by step. Safety analysis aims to identify risk factors that may endanger equipment, personnel safety, and normal production order during the ammonia synthesis process, providing early warnings and preventing safety accidents, including parameter threshold monitoring, failure mode identification, and safety trend prediction. Stability analysis focuses on the stability of various process parameters and production states during ammonia synthesis, ensuring continuous and stable production and avoiding product quality degradation or energy consumption increases due to production fluctuations, including parameter fluctuation analysis, system response analysis, and equipment operational stability assessment. Optimization analysis aims to improve the efficiency of ammonia synthesis production, reduce energy consumption, and enhance product quality. Through in-depth analysis of production data, it identifies methods and strategies for optimizing the production process, mainly including energy consumption optimization, product quality optimization, and production efficiency optimization.

[0086] In one possible implementation, this application proposes a safety index, stability index, and optimization index based on distance function determination. Its key feature is that it models historical data, sets a safety boundary for the mapped data, and calculates the current safety index based on the distance between the new mapping point and the boundary. The stability index is determined by the magnitude of variable fluctuations over a period of time, combined with the proportion of distance the new mapping point moves on the mapping surface. The optimization index is calculated based on the distance between the coordinate range representing the optimal operating condition in the clustered mapping points and the current position of the new mapping point. All three indices range from 0% to 100%, with higher indices indicating greater safety, stability, and proximity to the optimal operating condition. A low index triggers a safety warning, and an index of 0 triggers a safety alarm. Visualizing these indices improves understanding for engineers and large language models. It is worth noting that this application can also select other data-driven indices, methods, or criteria applicable to safety analysis, stability analysis, and optimization analysis, depending on the factory site conditions.

[0087] In one possible implementation, the Large Language Model (LLM) in this application can selectively call a large language model API in conjunction with a process flow knowledge base built on-site. This process flow knowledge base covers the ammonia synthesis production process, the plant's operating logs, and relevant regulations, helping the large language model reduce illusions when generating strategies and output reasonable and feasible strategies based on the actual situation of the plant. Simultaneously, two methods can be chosen: calling the API or local deployment. Calling the API is simpler and usually charges based on the output length (Token), resulting in lower costs; however, its confidentiality is weaker, requiring careful consideration if involving a large number of confidential technologies and documents. In this embodiment, the large language model only provides semantic interpretation and paraphrasing of the data model analysis and visualization results, and the relevant process materials are stored in a local knowledge base, posing no risk of leakage. Therefore, API calls are preferred. API calls require a continuous network connection, which can be limiting in scenarios with network restrictions. Local deployment, on the other hand, requires establishing a sufficient computing power cluster, which is costly. However, it eliminates the risk of data leakage and allows for fine-tuning of the large language model based on the established computing power cluster and relevant process materials, generating a large language model with factory-specific characteristics and improving the quality of the generated strategy. Alternatively, a large language model agent can be set up to generate strategies in a more professional tone. For example, the agent could be configured to speak in the tone of an "experienced chemical engineer," resulting in more accurate strategies and avoiding the misuse of technical terms. It is worth noting that the selection of the large language model in this application can be determined based on the actual situation of the factory; currently, there are various large oracle models available on the market.

[0088] It is worth mentioning that the scenarios described in the embodiments of this application are merely illustrative. The large language model embodiments in this application can be implemented in any scenario, including local offline, online, external interface, and cloud server. Local offline is suitable for enterprise groups, but the related hardware costs are relatively high. Online models allow direct interaction with the large model via a webpage, which is lower in cost but has poor scalability and flexibility. External interface mode connects to the large language model API, reducing related hardware costs, but its security is not as good as local deployment. Furthermore, using external interfaces also requires related token fees, i.e., the cost of the model API provider's computing resources consumed by the volume of input data into the large model. Cloud servers involve renting equipment or computing power from relevant intermediary vendors for downloading, connecting to, and calling the model. This method reduces hardware costs, but related data security issues require special attention. The selection of various large language models can be based on the scenario. Using the local offline method requires choosing an open-source large model or other large language models using external interfaces; the final choice of the base model can also be determined comprehensively based on enterprise restrictions, such as the requirement to use domestic large models and the enterprise's own budget.

[0089] For details, please refer to Figure 3 The diagram shown illustrates the generation process of a target optimization strategy according to an embodiment of this application. Figure 3Big data analysis was conducted on the boiler and shift conversion sections of ammonia synthesis production. Model calculations showed that the boiler section's safety index (Furnace) was 0%, triggering an alarm. The shift conversion section's safety index (Shift) was 8.27%, below the set warning threshold, triggering an alarm. According to the safety analysis, the fuel gas inlet flow measurement point FI_00001.PV in the boiler section had a value of 807.6 Nm³ / h, with a probability S(FI_00001.PV) of 95.39% causing an alarm. Similarly, according to the safety analysis, the furnace bed temperature measurement point TI_00002.PV in the shift conversion section had a value of 498.3℃, with a probability of 45.51% causing an alarm. Optimization analysis showed that the optimization index for point TI_00002.PV was 3.56%, indicating significant room for improvement, and the predicted probability O(FI_00001.PV) of this optimization potential was 67.03%. Specifically, the above safety probability calculation can be understood as the probability that the current point is the main cause of the section alarm or warning. The higher this probability, the more likely the point is to have a problem, and it should be investigated first. The optimization probability can be understood as the probability that there is room for optimization at the point. The higher the probability, the more likely the point should be considered in the optimization control strategy. Specifically, in this application embodiment, the analysis follows the principle of safety priority. That is, if a safety alarm occurs, the results of stability and optimization analysis are not considered as the main target input into the large language model, and the section alarm problem is solved first. In this application embodiment, if an alarm occurs in the boiler section, the relevant safety analysis results should be input into the large language model, and the other stability and optimization analysis results should not be input. Similarly, in this application embodiment, if a warning occurs in the transformation section, the safety analysis results should be processed first. Since there is a relatively significant optimization space at this point, the optimization analysis results are also input into the large language model. During the safe operation phase, the relevant results of stability analysis will be input into the large language model. Based on the above logic, this application can also select appropriate methods according to the factory's safety regulations.

[0090] Furthermore, as mentioned above Figure 3As shown, the results of the above safety analysis, stability analysis, and optimization analysis are input into a Large Language Model (LLM). Based on its understanding of the synthetic ammonia production process, causal reasoning, and multi-factor trade-offs, the LLM generates a control strategy. The boiler section's countermeasures include: triggering a boiler safety alarm and initiating on-site alerts; based on historical experience, and given the current high intake air volume, simultaneously reducing the boiler's induced draft and forced draft to prevent excessive carbon buildup and CO levels, and, if necessary, opening steam venting and depressurization; switching to a backup intake port to maintain a stable total fuel gas supply and closing the No. 1 intake port valve; checking the status of instruments and valves, troubleshooting fuel gas system problems, and confirming the control system logic; continuously recording and following up, and developing subsequent preventative measures. The measures taken for the shift conversion section include: triggering shift conversion early warnings and initiating early intervention; based on historical experience, given the current high temperature of the No. 2 furnace bed, implementing step-by-step steam control, adjusting the feed gas rate, and activating auxiliary furnace cooling measures; calibrating equipment and instruments, and conducting catalyst status checks; continuous prevention and increasing the frequency of inspections; recognizing that there is significant room for improvement in the target variable of steam consumption in the shift conversion section, optimization operations can be implemented after the early warning is canceled and normal operation is restored, adopting a step-by-step cooling and simultaneous steam reduction approach to reduce unnecessary steam consumption; and optimizing the feed gas adaptability to match CO concentration fluctuations. These strategies, through a large language model, convert the output of the big data model into easily understandable strategy statements, facilitating subsequent interpretation by on-site personnel. This enables effective monitoring and analysis of the boiler and shift conversion sections in the ammonia synthesis process, generating precise control strategies, ensuring production safety and stability, and providing direction for production optimization.

[0091] Step 105: Based on the target optimization strategy, optimize and control the ammonia synthesis production site.

[0092] In this embodiment of the application, after generating the target optimization strategy for the ammonia synthesis production process, the equipment at the ammonia synthesis production site can be optimized and controlled according to the optimization strategy, thereby improving the production efficiency of ammonia synthesis.

[0093] In one possible implementation, embodiments of this application can control the production equipment at the ammonia synthesis production site to perform corresponding operations through a target optimization strategy, and update the production modeling data based on the operation results, thereby achieving a closed loop of ammonia synthesis production control and optimization.

[0094] On the other hand, reference Figure 4 The image shows a synthetic ammonia production control system provided in an embodiment of this application. The system includes:

[0095] The data preprocessing module is used to acquire raw data from the ammonia synthesis field section and preprocess the raw data to obtain production modeling data.

[0096] The big data feature analysis module is used to perform unsupervised analysis on production modeling data based on a preset big data model, and to determine the key features and analysis results of the key features in the ammonia synthesis process.

[0097] The visualization module is used to visualize key features and generate visual charts.

[0098] The real-time analysis and strategy generation module is used to perform semantic parsing and causal reasoning on visual charts and analysis results based on a preset large language model, and generate target optimization strategies.

[0099] The control strategy output module is used to optimize the control of the ammonia synthesis production site based on the target optimization strategy.

[0100] Specifically, through the aforementioned ammonia synthesis production control system, this application can input various raw data from the ammonia synthesis site into the data preprocessing module. The data preprocessing module organizes, cleans, and standardizes this raw data. In this embodiment, the raw data includes unstructured or semi-structured data such as real-time flow, temperature, and pressure collected by sensors during the ammonia synthesis production process, as well as information from each point, such as the corresponding work section, upper and lower thresholds, and setpoints. In the organization stage, data from different sources and formats are categorized and integrated; the cleaning stage removes outliers caused by sensor drift and noise data such as instantaneous fluctuations during equipment start-up and shutdown; the standardization process unifies the data format, units, and acquisition frequency, ultimately obtaining structured production modeling data, laying the foundation for subsequent analysis.

[0101] Next, the production modeling data is transmitted to the Big Data Feature Analysis Module. This module performs unsupervised analysis on the production modeling data based on the big data model, extracting key features such as the temperature, pressure, and hydrogen-nitrogen ratio of the synthesis tower affecting ammonia production, and key process parameters affecting energy consumption such as coal consumption in the gasifier section and electricity consumption in the air separation section. Simultaneously, it constructs a correlation model between these key features based on the ammonia synthesis process mechanism to more accurately reflect the variable relationships in the production process. On one hand, the key features extracted by the Big Data Feature Analysis Module are transmitted to the Data Visualization Module; on the other hand, the analysis results from the Big Data Feature Analysis Module are also transmitted to the Real-Time Analysis and Strategy Generation Module. The visualization module performs multi-form visualization processing on the key features, such as displaying temperature changes using real-time trend curves and presenting load differences between sections using heat maps of operating conditions, making the production data more intuitive and understandable. The Real-Time Analysis and Strategy Generation Module performs real-time analysis of the on-site operating conditions based on the big data model, including anomaly identification of key features, safety risk assessment, and spatial calculation of energy consumption optimization. Meanwhile, based on a pre-set large language model, the visualized charts are interpreted, and the real-time analysis results are semantically parsed to jointly generate a target optimization strategy that includes logical reasoning and process constraint adaptation.

[0102] Furthermore, the target optimization strategy generated by the real-time analysis and strategy generation module is transmitted to the control strategy output module. Based on the target optimization strategy, the control strategy output module converts it into executable control commands, such as adjusting valve opening and modifying unit frequency, and directly sends them to the DCS system or PLC equipment at the ammonia synthesis production site to control the production process, achieve precise regulation under dynamic operating conditions, and reduce capacity fluctuations.

[0103] Specifically, in this embodiment, the raw data represents key parameters of the continuous ammonia synthesis process, including but not limited to: state variables (e.g., pressure, concentration, etc., reflecting system behavior), operational variables (e.g., temperature, flow rate, etc., directly controllable parameters), and constraint variables (e.g., temperature upper and lower limits, pressure thresholds, etc., safety boundary parameters). Production modeling data includes but is not limited to: section identifiers, real-time data values, sample collection time, and engineering units. Through the data preprocessing module, this embodiment ensures the completeness and standardization of factory production data, thus laying a solid foundation for subsequent modeling and analysis.

[0104] In one possible implementation, refer to Figure 5 The diagram shows a data preprocessing module provided in an embodiment of this application. In this embodiment, the raw data from the ammonia synthesis production site can come from a distributed control system (DCS). This application can aggregate and organize the data from each section according to a standard time frequency. For example, the raw data can be all data after the plant is officially started up and running, with a sampling frequency of once every 10 minutes. It is worth mentioning that the selection of the sampling frequency in this application can take into account the upper limit of the plant's data storage frequency and the existing computer hardware level on site. This embodiment does not make specific limitations on this. For example, in this embodiment, the data stored in the plant's DCS reaches the second level, but considering the multiple sections and multiple locations and the relatively congested allocation of the plant's computer facilities, this application can choose a sampling frequency of once every 10 minutes. At the same time, the time range can also be selected based on the plant's operating history. For example, the time period of trial operation or semi-operation, the time period before the new process is updated, and the time period when equipment or network failures cause problems with on-site data storage should be avoided as much as possible.

[0105] Next, according to the above Figure 5As shown, the Data Preprocessing Module performs a series of processing operations on the collected and organized raw data. These methods include, but are not limited to, downsampling, normalization, standardization, missing value handling, outlier analysis, noise identification, and removal of non-numerical values. Specifically, downsampling reduces data redundancy and improves subsequent processing efficiency; normalization and standardization transform data of different dimensions and ranges into a unified standard, facilitating analysis; missing value handling fills in gaps in the data, ensuring the integrity of time-series data; outlier analysis and noise identification identify and process abnormal data, preventing interference with subsequent modeling; and removing non-numerical values ​​removes irrelevant characters and other data, making the data more suitable for analysis. In particular, data cleaning and transformation operations should be selected based on the actual conditions of the data. In addition to the data processing methods listed above, other methods from various statistical fields can also be used, but over-processing should be avoided to prevent data loss. Specifically, in this embodiment, data from eight different sections of the ammonia synthesis plant are preprocessed to obtain eight sets of production modeling data for each section. Simultaneously, the raw data is preprocessed without section division to obtain one set of full-process production modeling data, resulting in nine sets of data ("one large and eight small"). It is worth noting that this application can group data according to different plant conditions, and the number of data groups is not specifically limited. Subsequently, the processed production modeling data is packaged, sent to the big data feature analysis module, and stored in a relevant database for backup. In this embodiment, the production modeling data can be stored in a database management system (My Structured Query Language, MySQL) for convenient subsequent data retrieval and model updates. Thus, through the aforementioned data preprocessing module, this application can transform the plant's complex raw data into high-quality basic data, creating an important prerequisite for accurate analysis and monitoring of ammonia synthesis production.

[0106] In one possible implementation, refer to Figure 6The diagram illustrates the workflow of a Big Data Feature Analysis Module provided in this embodiment. In this embodiment, the big data modeling process includes two main operations: the application of dimensionality reduction algorithms and model training. The trained big data model can analyze and process the input real-time data, outputting model results. The establishment of the big data model provides a basis for subsequent modules such as real-time analysis and strategy generation, enabling accurate analysis and monitoring of ammonia synthesis production. The model results include analysis results of key features in the ammonia synthesis production process, such as anomaly identification of various parameters, energy consumption analysis, and production prediction-related features. Next, the big data feature analysis model can transmit the analysis results and other information to the Data Visualization Module and the Real-Time Analysis and Strategy Generation Module, respectively, and this output process is continuously updated and repeated. In this embodiment, sampling can be performed every 10 minutes, meaning the relevant analysis results of the big data model are updated every 10 minutes and transmitted to the visualization module and the real-time analysis and strategy generation module every 10 minutes.

[0107] In one possible implementation, refer to Figure 7 The diagram illustrates the workflow of a visualization module provided in this embodiment. In this embodiment, the visualization module can visualize the big data model generated by the previous module in various forms, including but not limited to: multi-dimensional big data pattern diagrams, variable trend charts, and security analysis lists. Simultaneously, it updates the relevant charts based on continuously updated new data, achieving real-time visualization of factory production.

[0108] In one possible implementation, refer to Figure 8 The diagram shown is a flowchart of a real-time analysis and strategy generation module provided in an embodiment of this application. In this embodiment, security analysis, stability analysis, and optimization analysis can be performed on the model results from the big data feature analysis module, and the results of these analyses, along with visual charts, are input into the large language model. Thus, the large language model can combine process knowledge and historical experience from ammonia synthesis production to understand and reason about the input analysis results, and generate target optimization control strategies to guide the control operations at the ammonia synthesis production site, thereby achieving optimized, safe, and stable control of the production process.

[0109] In one possible implementation, such as Figure 9This is a schematic diagram of the control process of a target optimization strategy provided in an embodiment of this application. The control strategy output module receives a control strategy that includes event location, cause analysis and countermeasures. The central control room, as the core node for information processing and distribution, receives the control strategy. Based on the control strategy, the central control room, combined with the actual working conditions of ammonia synthesis production and the division of responsibilities of each position, transmits relevant information to the responsible positions, including workshop director, process engineer, safety engineer, equipment engineer and operator.

[0110] Specifically, the workshop director receives information from the central control room regarding overall production scheduling and resource coordination, enabling them to oversee the adjustment and optimization of ammonia synthesis production and coordinate between different work sections. The process engineer transmits control strategies related to process parameter adjustments and production process optimization, helping them to precisely adjust the ammonia synthesis process and ensure its rationality and efficiency. The safety engineer transmits information from control strategies regarding safety risk prevention and control, and the implementation of safety measures, enabling them to conduct safety inspections and risk assessments to ensure the safety of the ammonia synthesis production process. The equipment engineer transmits control strategies related to equipment operation status monitoring and maintenance, allowing them to promptly understand equipment conditions and perform necessary maintenance and repairs to ensure stable equipment operation. The operator transmits specific operational instructions, guiding operators on appropriate operations at the production site, such as valve opening adjustments and equipment start-up and shutdown, ensuring the implementation of control strategies on the production line.

[0111] It is worth mentioning that the above strategy allocation details can be adjusted according to the actual situation of the factory, and the transmission should be as accurate as possible. In this way, through the workflow of the above control strategy output module, the embodiment of this application realizes the complete process of control strategy from generation to transmission and execution by the responsible positions, ensuring that the ammonia synthesis production can be carried out efficiently, safely and stably according to the optimized control strategy. It achieves a perfect closed loop of the entire ammonia synthesis analysis and real-time monitoring system.

[0112] Specifically, in this application embodiment, the above-mentioned synthetic ammonia production system is applied to a 300,000-ton / year synthetic ammonia plant. This system can provide early warning of production risks 30 minutes in advance by real-time monitoring of parameters in the entire process of synthesis, air separation, boiler, purification, gasification, and slag water. The generated optimization strategies significantly reduce raw material coal consumption, effectively improve conversion rate, and increase annual benefits by more than 10 million yuan, providing strong support for enterprises to achieve safe, efficient, and low-consumption production.

[0113] Please see Figure 10As shown, based on the same technical concept, this application also provides a computer device 100. In one embodiment, the computer device can be a device specifically for controlling the production of synthetic ammonia, or it can be a device for overall control of industrial production. The computer device, as shown... Figure 10 As shown, it includes a memory 1001, a communication module 1003, and one or more processors 1002.

[0114] The memory 1001 is used to store computer programs executed by the processor 1002. The memory 1001 may mainly include a program storage area and a data storage area. The program storage area may store the operating system and programs required to run instant messaging functions, etc.; the data storage area may store various instant messaging information and operation instruction sets, etc.

[0115] Memory 1001 may be volatile memory, such as random-access memory (RAM); memory 1001 may also be non-volatile memory, such as read-only memory, flash memory, hard disk drive (HDD), or solid-state drive (SSD); or memory 1001 may be any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto. Memory 1001 may be a combination of the above-described memories.

[0116] The processor 1002 may include one or more central processing units (CPUs) or digital processing units, etc. The processor 1002 is used to implement the aforementioned ammonia synthesis production control method when it calls the computer program stored in the memory 1001.

[0117] The communication module 1003 is used to communicate with the industrial control system.

[0118] This application embodiment does not limit the specific connection medium between the memory 1001, communication module 1003, and processor 1002. This application embodiment... Figure 10 The memory 1001 and the processor 1002 are connected via a bus 10010, and the bus 10010 is in Figure 10 The diagram uses thick lines to describe the connections between other components; these are for illustrative purposes only and should not be considered limiting. The 10010 bus can be divided into address bus, data bus, control bus, etc. For ease of description, Figure 10It is described using only a thick line, but does not indicate that there is only one bus or one type of bus.

[0119] The memory 1001 stores a computer storage medium, which stores computer-executable instructions. The computer-executable instructions are used to implement the synthetic ammonia production control method of the embodiments of this application, and the processor 1002 is used to execute the synthetic ammonia production control method of the above embodiments.

[0120] Based on the same inventive concept, embodiments of this application also provide a storage medium storing a computer program that, when run on a computer, causes the computer to execute the steps in the synthetic ammonia production control method according to various exemplary embodiments of this application described above.

[0121] In some possible implementations, various aspects of the synthetic ammonia production control method provided in this application can also be implemented in the form of a computer program product, which includes a computer program that, when run on a computer device, causes the computer device to perform the steps in the synthetic ammonia production control method according to various exemplary embodiments of this application described above. For example, the computer device can perform the steps of the various embodiments.

[0122] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0123] The program product of the embodiments of this application may employ a portable compact disc read-only memory (CD-ROM) and include a computer program, and may run on a computer device. However, the program product of this application is not limited thereto. In this application, the readable storage medium may be any tangible medium that contains or stores a program, and the computer program included therein may be used by or in conjunction with a command execution system, apparatus, or device.

[0124] A readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a readable computer program. This propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with a command execution system, apparatus, or device.

[0125] Computer programs contained on readable media may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0126] Computer programs for performing the operations of this application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages.

[0127] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0128] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0129] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0130] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0131] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for controlling ammonia synthesis production, characterized in that, The method includes: Acquire production modeling data for the ammonia synthesis site, wherein the production modeling data characterizes the structured process parameters and site configuration information during the ammonia synthesis production process; Based on a pre-defined big data model, unsupervised analysis is performed on the production modeling data to determine the key features in the ammonia synthesis process and the analysis results of the key features; the key features characterize the parameters affecting ammonia production and energy consumption, and the analysis results characterize the anomaly identification, energy consumption analysis, and production prediction results of each parameter; The key features are visualized to generate visualization charts; the visualization charts represent the operating status and working condition boundaries of the ammonia synthesis production process. Based on a pre-defined large language model, semantic parsing and causal reasoning are performed on the visualization charts and analysis results to generate target optimization strategies. Based on the aforementioned target optimization strategy, the synthetic ammonia production site is optimized and controlled.

2. The method as described in claim 1, characterized in that, The acquisition of production modeling data for the ammonia synthesis field section includes: Obtain the raw data of the ammonia synthesis field section, and identify and correct any abnormal missing values ​​in the raw data; The processed raw data is then standardized to obtain raw data with a uniform format. Based on a preset data integration strategy, the multi-segment data in the original data are correlated and integrated to obtain structured production modeling data.

3. The method as described in claim 1, characterized in that, The aforementioned unsupervised analysis, based on a pre-defined big data model, of the production modeling data determines the key characteristics of the ammonia synthesis process and the analysis results of these key characteristics, including: Based on a preset unsupervised clustering strategy, pattern recognition is performed on the production modeling data to obtain production datasets corresponding to multiple production condition categories. Based on a preset feature selection strategy, relevant key features are selected from each production dataset, and each key feature is transformed based on a preset dimensionality reduction strategy to obtain the dimensionality-reduced key features. Based on a preset time series analysis strategy, the key features after dimensionality reduction are subjected to pattern mining to generate corresponding feature evolution curves; the feature evolution curves characterize the dynamic changes of the corresponding key features with the production process.

4. The method as described in claim 1, characterized in that, The big data model is constructed through the following steps: Based on a preset dimensionality reduction strategy, the production modeling data is dimensionality reduced to obtain a training dataset. Based on a preset unsupervised learning strategy and time series analysis strategy, the model is trained on the training dataset to obtain the big data model.

5. The method as described in claim 1, characterized in that, The visualization charts include at least one of the following: a big data multidimensional pattern diagram, a variable trend diagram, and a safety analysis list; the big data multidimensional pattern diagram represents the collaborative relationship and safety boundaries of the operating status of each section in the ammonia synthesis production process. The variable trend graph represents the fluctuation pattern of each production parameter over time during the ammonia synthesis process; the safety analysis list includes quantitative safety information and abnormal early warning information of the ammonia synthesis production status.

6. The method as described in claim 1, characterized in that, The optimization and control of the ammonia synthesis production site based on the target optimization strategy includes: Based on the aforementioned target optimization strategy, the production equipment at the ammonia synthesis production site is controlled to perform corresponding operations. The production modeling data is updated based on the operation results.

7. A synthetic ammonia production control system, characterized in that, The system includes: The data preprocessing module is used to acquire raw data from the ammonia synthesis field section and preprocess the raw data to obtain production modeling data; the production modeling data represents the structured process parameters and section configuration information in the ammonia synthesis production process. The big data feature analysis module is used to perform unsupervised analysis on the production modeling data based on a preset big data model, to determine the key features in the ammonia synthesis process and the analysis results of the key features; the key features characterize the parameters affecting ammonia production and energy consumption, and the analysis results characterize the anomaly identification, energy consumption analysis and production prediction results of each parameter; The visualization module is used to visualize the key features and generate visualization charts; the visualization charts represent the operating status and operating condition boundaries of the ammonia synthesis production process. The real-time analysis and strategy generation module is used to perform semantic parsing and causal reasoning on the visualization charts and analysis results based on a preset large language model, and generate target optimization strategies. The control strategy output module is used to optimize and control the ammonia synthesis production site based on the target optimization strategy.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer storage medium storing computer program instructions thereon, characterized in that, When executed by a processor, the computer program instructions implement the steps of the method according to any one of claims 1 to 6.

10. A computer program product comprising computer program instructions, characterized in that, When executed by a processor, the computer program instructions implement the steps of the method according to any one of claims 1 to 6.

Images